Wind speed is an important meteorological condition affecting the urban environment. Thus, analyzing the typical characteristics of the wind speed diurnal variation is helpful for forecasting pollutant diffusion. Based on the K-means clustering method, the diurnal variation characteristics of the wind speed in Beijing during 2008–2017 are studied, and the spatio-temporal characteristics of the wind speed diurnal variations are analyzed. The results show that there are mainly 5–7 clusters of typical characteristics of the wind speed diurnal variation at different stations in Beijing, and the number of clusters near the city is smaller than that in the suburbs. The typical number of the wind speed diurnal variation during 2013–2015 is smaller than that in other periods, which means the anomalous clusters of the diurnal variation are reduced. Besides, the numbers of different clusters in different years are often switched. Especially, the switch between clusters 5 and 6 as well as the switch between clusters 6 and 7 are frequent. Based on the second cluster analysis of the clustering results at Beijing station, we find 12 clusters of the diurnal variation, including 9 clusters of “large in the daytime, while small at night”, 2 clusters of “monotonous”, and 1 cluster of “strong wind”. Furthermore, the low-speed clusters of wind mainly locate in the city with a significant increasing trend, while the high-speed clusters and the monotonous clusters of wind locate in the suburbs with a decreasing trend.

Diurnal variation of wind speed, Typical wind modes, K-means, clustering method